EFFECT OF GAPPING ON THE TRANSIENT AND SUSTAINED …

725

i) Principal Geotechnical Engineer, AECOM Australia Pty Ltd (formerly PhD Student, Centre for OŠshore Foundations Systems, The Univer-sity of Western Australia), Australia.

ii) Professor, Centre for OŠshore Foundation Systems, The University of Western Australia (susan＠civil.uwa.edv.au).iii) Winthrop Professor, ditto.

The manuscript for this paper was received for review on January 8, 2009; approved on August 5, 2010.Written discussions on this paper should be submitted before May 1, 2011 to the Japanese Geotechnical Society, 4-38-2, Sengoku, Bunkyo-ku,Tokyo 112-0011, Japan. Upon request the closing date may be extended one month.

725

SOILS AND FOUNDATIONS Vol. 50, No. 5, 725–735, Oct. 2010Japanese Geotechnical Society

EFFECT OF GAPPING ON THE TRANSIENT AND SUSTAINED UPLIFTCAPACITY OF A SHALLOW SKIRTED FOUNDATION IN CLAY

HUGO E. ACOSTA-MARTINEZi), SUSAN M. GOURVENECii) and MARK F. RANDOLPHiii)

ABSTRACT

Shallow skirted foundations are an attractive solution for supporting oŠshore platforms that are subject to upliftdue to overturning or buoyancy loading. In practice, gapping may occur along the skirt-soil interface leading to adetrimental eŠect on uplift capacity. This paper presents results from beam centrifuge tests that investigated the eŠectof gapping on the transient and sustained uplift capacity of shallow skirted foundations. The results indicate that thetransient uplift capacity following the formation of a gap was around 60z of that for intact soil contact along theskirt-soil interface. Under sustained uplift, the time to accumulate a displacement of 1z of the foundation diameterwas reduced by an order of magnitude due to the presence of a gap.

Key words: bearing capacity, clay, gapping, model test, skirted foundation, uplift (IGC: E3/H3)

INTRODUCTION

Skirted foundations comprise a base plate with aperipheral skirt, and sometimes an arrangement of inter-nal skirts, that penetrate the seabed conˆning a soil plug.Foundation skirts oŠer various beneˆts: transmittingfoundation loads to deeper, stronger soil leading to in-creased capacity and reduced displacements, contributingitself to the augmented capacity, providing protectionagainst scour, and accommodating variations in seabedproˆle. A particular beneˆt of skirted foundations is theirenhanced uplift resistance due to negative excess porepressures generated in the soil plug during overturning oruplift. While these negative excess pore pressures, or pas-sive suctions, can be relied on, uplift resistance isgoverned by a reverse end bearing mechanism. With time,water will ‰ow into the soil plug, the suctions will dissi-pate and the uplift resistance will reduce to the frictionalresistance mobilised along the foundation skirts. The en-hanced capacity of skirted foundations to resist uplift,due to passive suctions developed within the soil plug,makes them particularly attractive foundation systemsfor oŠshore facilities that experience overturning or up-lift.

However, oŠshore design guidelines for shallow foun-dations are based on classical bearing capacity theoriesthat do not account for enhanced uplift resistance due topassive suctions (Det Norske Veritas (DNV), 1992;American Petroleum Institute (API), 2000; InternationalOrganization for Standardization (ISO), 2003). Passive

suctions have been relied on in practice in the design ofsome skirted foundation systems, but in each case theproject was based on extensive êld, experimental andanalytical studies (Støve et al., 1992; Dyvik et al., 1993;Tjelta and Haaland, 1993; Bye et al., 1995). A better un-derstanding of the development and dissipation of pas-sive suctions within skirted foundations would lead toe‹ciencies in design of current skirted foundation sys-tems in terms of reducing foundation size and extendingthe period over which passive suctions can be relied on.Interest also exists in developing a new concept ofoŠshore storage facility that would be buoyant followingoŒoading for a period of a few days up to several weeks.

Recent experimental centrifuge studies at the Centrefor OŠshore Foundation Systems (COFS) at The Univer-sity of Western Australia have considered the transientand sustained uplift capacity of shallow skirted founda-tions under concentric, monotonic and cyclic loading andhave shown encouraging results for skirt depth to foun-dation diameter ratios d/D as low as 0.3 (Gourvenec etal., 2007; Acosta-Martinez and Gourvenec, 2008;Acosta-Martinez et al., 2008; Gourvenec et al., 2008).The previous experiments considered intact contact be-tween the external skirt walls and the soil. In reality, gap-ping may occur along the skirt-soil interface due to theharsh environmental forces, leading to a reduction in thedrainage path length, accelerating the dissipation of pas-sive suctions within the soil plug. Once a gap has devel-oped, it can propagate with time due to repeated loadsfrom wind, waves and currents. The deeper the extent of

726

Fig. 1. Foundation model and location of sensors

726 ACOSTA-MARTINEZ ET AL.

the gap, the higher the stress concentration induced by itspresence in the soil mass that remains in contact with thefoundation. Gapping can lead to softening of the materi-al and local reduction of shear strength as well as loss ofsoil in the vicinity of the gap due to pumping scour insideand around the cavity.

The likelihood of gap formation and propagation de-pends on the loading regime and soil conditions.Although no literature is available regarding gappingalong shallow skirted foundations, various authors havereported that large displacements are required for a gapto form along laterally loaded caissons (similar to skirtedfoundations but with larger embedment depth to di-ameter ratio, d/D) in normally consolidated soil, whilefor lightly and over consolidated soils a gap may appearimmediately after lateral or inclined loading (Clukey etal., 1995; Randolph et al., 1998; Clukey et al., 2003;CoŠman et al., 2004). Clukey et al. (1995) report resultsfrom a centrifuge study of the eŠect of gapping on thecapacity of suction caissons in typical Gulf of Mexiconormally consolidated soft clay under cyclic loading, butdid not observe a critical load at which the caisson sud-denly failed from gapping. Randolph and House (2002)considered the sustained uplift capacity of suction cais-sons through analytical studies validated against beamcentrifuge tests in normally consolidated clay and recog-nised the possibility of a hydraulic leak down the side ofthe caisson, due to slight tilting during cyclic loading,that may allow tensile fracture at the caisson base. Forthis case, they proposed an intermediate mode of failure,between pure shearing along the skirt and the full reverseend bearing. CoŠman et al. (2004) investigated thehorizontal capacity of suction caissons in normally con-solidated clay at 1 g and did not observe a distinct gap inthe back side of the caisson after lateral displacements upto several times the caisson diameter, but observed ascarp on the back side of the caisson and a wedge of soilthat slid downwards, maintaining contact with the cais-son. Supachawarote et al. (2005) reported three-dimen-sional ˆnite element analysis investigating the eŠect ofgapping on the inclined pullout capacity of suction cais-sons and for an embedment ratio d/D＝1.5 showed anaverage reduction in inclined undrained pullout capacityof 27z in lightly over consolidated soil.

The eŠect of gaping along the interface of shallowskirted foundations has not previously been considered,although it is arguably more critical than for suction cais-sons, because of shorter drainage path lengths leading tomore rapid dissipation of passive suctions from withinthe soil plug.

This paper presents results of beam centrifuge tests ona circular skirted foundation in which a gap was deliber-ately created on one side of the foundation in some of thetests. The study evaluates the eŠects of the gap on thefoundation response under transient and sustained uplift,comparing results for tests with intact soil contact alongthe skirt-soil interface, and tests where a gap was createdprior to loading the foundation.

EXPERIMENTAL SET-UP

Physical modelling was carried out in the 1.8 m radiusbeam centrifuge at COFS (Randolph et al., 1991) at anacceleration of 167 g. A swinging platform at a radius of1.8 m provides a nominal working radius of 1.55 m; thecentrifuge has a maximum payload of 200 kg at the maxi-mum acceleration of 200 g, and is thus rated at 40 g-tonnes. The basic principle of centrifuge modelling is thatthe self-weight stresses in the model are enhanced by thecentrifugal acceleration in order to give stresses (andshear strengths) that are homologous in model and proto-type. Essentially, stresses and strains scale 1:1 betweenmodel and prototype. In centrifuge modelling, all lineardimensions of the model are scaled by 1:N relative to theprototype, and a centrifugal acceleration of N timesearth's gravity (g) is applied during the test, where N iscalled the scaling ratio. A detailed description of scalinglaws in geotechnical centrifuge modelling can be found el-sewhere (Taylor, 1995; Muir Wood, 2004; Garnier et al.,2007).

Tests without gapping, used as a reference to compareagainst tests with gapping, were carried out with a con-ventional one-dimensional actuator, while a special ar-rangement including a second linked actuator was devel-oped for the tests involving gap generation (as describedlater).

Foundation Model and InstrumentationFigure 1 shows a schematic of the foundation model

and instrumentation. The model has a foundation di-ameter Dm＝120 mm and skirt depth dm＝36 mm, corre-sponding to prototype dimensions Dp＝20 m dp＝6 m atthe design acceleration level of the tests. A foundationmodel with a skirt depth to diameter ratio of d/D＝0.3 isconsidered as previous tests have shown reverse end bear-

727

Fig. 2. Foundation and T-bar tests layout

Fig. 3. Typical T-bar results: (a) Undrained shear strength proˆle and(b) Ratio of extraction to penetration resistance

727GAPPING EFFECT ON UPLIFT CAPACITY

ing can be mobilised. Similar foundations with lower em-bedment ratios (d/D) have been considered in previoustests but failed with a local shear mechanism (Gourvenecet al., 2008).

The foundation is equipped with a peripheral skirt andan internal cruciform stiŠener creating four compart-ments. The base plate is machined from aluminium andthe skirts and stiŠeners from stainless steel in order tominimise the foundation weight while ensuring su‹cientstiŠness of the skirt and stiŠeners. The foundation baseplate is equipped with a drainage vent that remains openduring installation and is closed with a plug cap to sealthe foundation during the tests. A drainage hole is locat-ed at the intersection of the cruciform on the underside ofthe base plate to allow water to pass between the skirtcompartments allowing drainage through the vent in thebase plate during installation.

The foundation is equipped with total pressure andpore pressure transducers (TPTs and PPTs, respectively).A TPT and a PPT are located on the underside of thebase plate to monitor variations of total pressure andpore pressure inside the foundation, assess contact of thesoil plug with the base plate during installation, trackvariation of suctions during transient and sustained up-lift, and determine separation during uplift. Additionalpairs of total pressure transducers are provided at di-ametrically opposed sections at two levels along theperipheral skirt, ‰ush with the external face. These sen-sors allow measurement of radial stresses and indicateseparation along the skirt-soil interface in tests involvinggapping and allow veriˆcation about verticality of themodel during installation and uplift. Two PPTs are ˆxedto the internal stiŠener at skirt tip level within a protectivehousing. The intention of these sensors was to monitorthe pore pressure variations at foundation level but theirresponse is aŠected by local bearing capacity failurearound the housing rather than re‰ecting the globalresponse of the foundation. Results from the PPTs atskirt tip level are therefore not presented.

Soil SampleA lightly over consolidated clay sample was prepared

from commercially available kaolin powder that is usedroutinely for centrifuge testing at COFS. The basic geo-technical properties of the kaolin used are: plasticity in-dex Ip＝34, speciˆc gravity Gs＝2.6 and compression in-dex Cc＝0.476. Dry kaolin powder was mixed under vacu-um with distilled water at a ‰uid content of 120z, corre-sponding to twice the liquid limit. The slurry was trans-ferred to a high capacity consolidation press, applyinggradual increments of load up to 150 kPa over 14 days,followed by gradual unloading to 40 kPa (the expectedeŠective vertical stress at skirt tip level) to obtain a lightlyover consolidated sample. The sample was then recon-solidated in the centrifuge at test acceleration level, 167 g,for 65 hours (corresponding to more than 200 years atprototype scale). The consolidation history led to an insitu void ratio e＝1.3, wet unit weight gt＝17 kN/m3 andover consolidation ratio (OCR) at skirt tip level of 3.8.

The undrained shear strength of the sample was deter-mined in-‰ight by T-bar penetrometer tests (Stewart andRandolph, 1991) carried out across the sample, beforeand during the programme of foundation tests. The loca-tion of the T-bar tests is presented in Fig. 2 alongside thefoundation test layout. The T-bar tests were carried outat a rate, [＝1 mm/s to ensure undrained conditions. Interms of the dimensionless velocity group V＝[L/cv,(where L is an appropriate length dimension of thedrainage path (in this case equivalent to the T-bar di-ameter, DT-bar＝0.005 m) and cv is a representativecoe‹cient of consolidation taken as 2.6 m2/year for theeŠective vertical stress at skirt tip level, [＝1 mm/s givesa dimensionless velocity V＝60, which exceeds the valuefor which undrained behaviour is achieved (Finnie andRandolph, 1994; Randolph and Hope, 2004). Typicalproˆles of undrained shear strength from the T-bar tests,using a constant T-bar factor NT-bar＝10.5 (Stewart andRandolph, 1994), and the ratio of extraction to penetra-tion resistance are presented in Fig. 3.

Gap GenerationOne of the technical challenges for this series of tests

was to generate a gap in-‰ight, to avoid a temporaryramp down of the centrifuge that would aŠect the stresshistory of the soil, the magnitude of pore pressure inside

728

Fig. 4. Setup for in-‰ight gap generation in the beam centrifuge: (a)scheme of actuators and load cells; (b) actual arrangement for testT4


the soil plug and radial stresses against the skirt. An ar-rangement that was developed to enable the formation ofa gap in the centrifuge, in-‰ight is illustrated in Fig. 4.The system employs two actuators, three load cells and achain linking the two actuators through a pulley system.Both actuators sit on top of the strongbox and permitvertical and horizontal displacement and loading of themodel.

Actuator #2 moves vertically upwards at a constant dis-placement rate of 0.05 mm/s, applying a tension to thechain leading to translation of the foundation and actua-tor #1 and forming a gap on the trailing edge of the foun-dation. A normalised lateral displacement wH/D¿5zwas found to provide a stable gap, with reduced (active)pressures on the trailing edge and development of high(passive) normal stresses on the leading edge. The lateralloading point of the foundation is at the level of the baseplate, so that some (albeit very limited) rotation accom-panies the horizontal translation of the foundation, lift-ing the trailing edge slightly. A maximum rotation of1/200 was calculated from variation in embedment dur-ing the gap generation. Vertical loads on the foundationare monitored with a ±3 kN load cell (V), bending of theloading column in actuator #1 is measured with a com-bined VM load cell, and the variation of the horizontalload (H) on the foundation is monitored with a ±8 kNload cell connected between the foundation model andthe chain. The V and VM load cells are connected inseries, with VM at a lower level where bending is morecritical as the loading arm is clamped at two positions in

the upper portion ( see Fig. 4(b)). A bearing was placedabove the VM load cell as an additional safety measure toallow slight rotation of the foundation while minimisingapplication of moment to the foundation.

Initially, the horizontal loading chain is left with mini-mum slack. An increase in the readings from the horizon-tal load cell is an indication of tightening of the chain andthe beginning of load application on the foundation.Bending is measured by the VM load cell and used to pro-gram a feedback loop in the software control to translatethe main actuator (#1) to keep bending in the foundationand loading arm close to zero. The feedback loop isnecessary in order to maintain the verticality of the mainactuator and subsequently concentric loading on thefoundation, and to avoid damaging either the loadingarm or the vertical load cell. Conventional uniaxial loadcells are not designed to support bending or momentloading and are fragile in that respect, both physicallyand in the accuracy of the readings when subjected tocombined loading. The arrangement of two load cells inseries ensured close verticality of the loading arm and reli-ability of the vertical load measurements. The actuatorsare controlled by an interface software written in-housebased on National Instruments' LabVIEW.

Extensive tuning of the proportional-integral-deriva-tive (PID) controller was required to adjust the sensitivityof the feedback loop, reducing the oŠset between the re-quired and measured horizontal displacement on theskirted foundation. A critical aspect was to avoid bend-ing in the opposite direction of that congruent with theapplication of the horizontal load, as this will induce areversal in the horizontal displacement and in practice arapid series of lateral cyclic displacements that willproduce gapping, softening and eventual remoulding andactive conditions on both sides of the foundation.

Vertical displacements of the two actuators are directlyrecorded by encoders placed on the motor of each actua-tor. In both cases the vertical displacement is monitoredby a linear displacement transducer (LDT) with 25 mmstroke and 1 mm resolution, installed on the frame towhich the actuator is connected. These devices are not indirect contact with either the foundation or the chain(pulley system).

Each gap was created at a rate of 0.05 mm/s, taking120 seconds in the centrifuge, corresponding to 39 days atêld scale. The rate of gap generation was chosen to allowprecise control of the actuators and minimise the genera-tion of excess pore pressures inside the soil plug ratherthan to replicate a particular êld process. In reality, gap-ping may occur over diŠerent time scales, ranging fromrapid, from extreme environmental loads to gradual ac-cumulation of lateral displacement or rotation underoperational loading conditions. The purpose of theprogramme of centrifuge tests reported in this paper wasto evaluate the eŠect of a gap between the soil and theskirt on the transient and sustained uplift resistance ofthe foundation, and not to model the process of gapgeneration or particular load paths that could generategapping.

729

Table 1. Testing programme

Reference Test sequence

T1 Undrained uplift (qu) (no gap)

T2_1 Sustained uplift: q/qu＝0.22 (no gap)æ intermediate consolidation (sc)

T2_2 Sustained uplift: q/qu＝0.7 (no gap)

T3_1 Gap formation, tc＝0 & undrained uplift (qu gap)æ intermediate consolidation (sc)

T3_2 Gap formation, tc＝0 & sustained uplift q/qu gap＝0.22æ intermediate consolidation (sc)

T3_3 Gap formation, tc＝0 & sustained uplift q/qu gap＝0.7æ intermediate consolidation (sc)

T3_4 Undrained uplift

T4_1 Gap formation and consolidation (sc)Undrained uplift (qu gap*) followed by push-back to sc

T4_2 Gap formation and consolidation (sc)Sustained uplift: q/0qu gap*¿0.22 followed by push-backto sc

T4_3 Gap formation and consolidation (sc)Sustained uplift: q/qu gap*¿0.7 followed by push-backto sc

T5 Undrained uplift (qu) (no gap)

T＝test stage reference; qu, qu gap and qu gap* are undrained upliftresistances for ULS.


TESTING PROGRAMME

The experimental program included a total of ˆve tests,referenced as T1 to T5, summarised in Table 1. Tests T1,T2 and T5 evaluated the transient and sustained upliftresponse of the foundation without gapping while testsT3 and T4 considered uplift resistance with gapping. Theundrained uplift capacity, qu, was determined in test T1.In the second test, T2, the foundation behaviour undertwo levels of sustained uplift loading, q/qu¿0.22 and 0.7(expressed as a fraction of qu assessed in T1) was investi-gated. The eŠect of gapping on the transient upliftresponse was then evaluated immediately after formationof a gap (in Test T3), and following consolidation aftergap generation (in Test T4). The series of tests was con-cluded with veriˆcation of the undrained uplift capacitywithout gapping (T5). Typical sketches of time historiesduring installation, consolidation, undrained uplift andsustained uplift for the same model and soil conditionshave been presented in Acosta-Martinez et al. (2008).

Due to the space restrictions due to the setup with thetwo actuators, only two locations were available for thetests with gapping, as seen in Fig. 2. As a result, un-drained and sustained load tests were carried out at thesame location in both Tests T3 and T4. In between teststhe foundation was pushed back by reloading to the con-solidation stress level, sc (explained below), and substan-tial primary consolidation was permitted before a newgap was formed for the next test. Although considerabledissipation of excess pore pressures generated duringreloading of the foundation was allowed for betweentests, cumulative damage will have occurred at the site ac-companied by some loss of embedment (since reloading

to the consolidation stress level rarely gave complete re-penetration of the foundation skirts). To minimise thiseŠect in the evaluation of sustained uplift response, thelower level of sustained uplift was evaluated ˆrst in allcases.

All tests followed vented, jacked in-‰ight installationat a constant rate of displacement of [＝0.1 mm/s, corre-sponding to a normalised velocity V＝[D/cv¿145, whereD is the diameter the foundation and cv is an averagevalue for the coe‹cient of consolidation of the kaolinclay, taken as 2.6 m2/year (Acosta-Martinez and Gour-venec, 2006). The rate of installation is su‹ciently rapidto ensure undrained conditions with respect to the basearea of the foundation, but local drainage in the vicinityof the skirts is likely to have taken place. The displace-ment rate during installation was selected to maintainprecise control of the centrifuge and not to represent theduration of êld installation, which usually takes a fewhours and would be impractical to replicate in the cen-trifuge.

Installation was considered complete when the baseplate made contact with the soil surface, and the stresslevel at that instance is deˆned by the installation stress,si. Following installation, a further stress increment wasapplied to a pre-determined consolidation stress level, sc.The consolidation stress, sc, was selected as twice the in-stallation stress, si. Following installation and applica-tion of the additional stress increment, the drainage ventwas sealed and the foundation was held at the consolida-tion stress level, sc, for a period of time, tc, to allow sub-stantial dissipation of the excess pore pressures generatedduring installation, to ensure a similar starting point(eŠective stress) in all tests. A detailed analysis of theseconsolidation stages under similar conditions has beenpresented by Acosta-Martinez and Gourvenec (2010) andindicates an average degree of consolidation of U＝90ztook place during the intermediate re-consolidation fol-lowing installation and prior to gapping or uplift.

EXPERIMENTAL RESULTS

InstallationThe stress-displacement response during installation

and initial consolidation was repeatable in all tests.Figure 5 shows the measured stress-displacementresponse during installation and consolidation during testT1 as an example. The stress-displacement response dur-ing the subsequent undrained uplift is also shown forcompleteness of the test sequence.

Installation resistance is expressed in terms of anaverage (vertical) installation stress, q, given by the verti-cal load cell reading divided by the cross-sectional (bear-ing) area of the foundation (q＝Q/A). Touch down isidentiêd from the readings of the TPT and PPT underthe base plate. The total installation resistance (si) at themoment of touch down is ¿28 kPa, indicating an inter-face friction ratio a＝0.34 based on a simple analyticalprediction using a bearing capacity factor for tipresistance Nc(tip)＝7.5. Assuming an average value of un-

730

Fig. 5. Stress history during test T1 Fig. 6. Synchronized movement of two actuators during in-‰ight gapgeneration

Fig. 7. Radial stresses variation during ˆrst gap formation

Fig. 8. Radial stress variation during gap formation after initial activegap


drained shear strength, su, over the depth of the skirt of13.5 kPa gives a ratio of installation resistance to un-drained shear strength, si/su＝2.0. Penetration of be-tween 90z and 95z of the depth of the skirts wasachieved when full contact was made between the baseplate and the soil due to heave of the soil plug within theskirt compartment.

Following touch-down of the base plate an additionalstress increment was applied to ensure good contactacross the base plate, sc. A period of consolidation atconstant stress was then permitted allowing substantialdissipation of excess pore pressures generated during in-stallation. The consolidation stress was deˆned as sc¿2si. All the tests followed the same installation and con-solidation procedure.

Gap GenerationFigure 6 shows a typical time history of vertical dis-

placement of the chain-pulley system and the resultinghorizontal displacement of the foundation during forma-tion of a gap. The chain moved at a constant displace-ment rate and the horizontal displacement of the founda-tion started while the initial slack of the chain was beingtaken up and before it became taut. EŠectiveness of thefeedback loop between the two actuators allowing bend-ing only in one direction is observed from the similaritybetween the displacement rates of the two actuators andthe absence of reverse horizontal movements on the foun-dation. However, it should be noted that the àpparent'horizontal displacement rate of the foundation is higherthan that of the chain at the start of the process, whereslight rotation may have occurred. This is also related tothe gradual development of the passive resistance on theleading edge of the foundation, giving rise to a small mo-ment on the foundation. In the ˆnal portion of the gapgeneration, when the chain is fully taut, the displace-ments rates are equivalent.

Figure 7 shows the response of the total pressure trans-ducers (TPTs) along the skirt during gap formation in-dicating the mode of foundation displacement throughchanges in radial stresses with respect to the values justprior to installation. The reduction in the radial stress onthe active side (Fig. 7(a)) and the increase in radial stress-

es on the passive side (Fig. 7(b)) is clear. As the requiredmovement to reach an active condition is several ordersof magnitude lower than for the corresponding passivecondition, there is a relatively stable reading in a portionof the plot on the active side (TPT#2) while radial stressescontinue to increase on the passive side. A normaliseddisplacement wH/D¿5.7z was applied for this particu-lar stage. Some transient signal disturbance in TPT#3 isobserved towards the end of the test, including negative

731

Fig. 9. Transient (undrained) uplift resistance

Fig. 10. Pore pressure generation beneath base plate during transientuplift


values that suggest transient suction prior to separationfrom the soil, with the gap ˆlling with water.

Due to the space limitations for the tests with gapping,multiple tests were carried out at each site. The founda-tion was pushed-back after each stage and consolidationwas permitted between tests and a new gap formed priorto each test. Figure 8 shows the variation of radial stress-es during formation of a `second' gap. This followed anormalised horizontal displacement of wH/D¿5.1z. Itcan be inferred that the initial gap did not reach the fullskirt depth, or some closure occurred, as there is an addi-tional decrease in the radial stress during formation ofthe second gap measured by the TPT at the lower level onthe active side (TPT#3) while the reading of the TPT lo-cated on the upper level on the active side (TPT#2)remains unchanged. The increase in radial stress atTPT#3 location before gap formation following an initialactive gap is due to the intermediate push-back and con-solidation between stages.

Undrained CapacityFigure 9 shows the undrained uplift resistance meas-

ured during each of the tests in terms of the average verti-cal uplift stress, q (＝load cell reading divided by crosssectional area of the foundation), and normalized upliftdisplacement, w/D, where w represents the total verticaldisplacement and D the foundation diameter. Undraineduplift capacity qu＝－150 kPa was measured in testswithout gapping (tests T1 and T5), mobilised at a nor-malised displacement w/D¿7–8z. Taking the un-drained shear strength at the tip level of the foundation assu(tip)¿18 kPa (from Fig. 3), gives a bearing capacity fac-tor Nc＝qu/su(tip)＝8.3. A simple mechanism of shearingalong the skirts and internal stiŠeners gives an upliftcapacity of ¿－18 kPa, taking an average shear strengthover the depth of the skirts of 13.6 kPa, indicatingmobilisation of reverse end bearing.

The undrained uplift capacity immediately followingformation of a gap (Test T3_1) qu gap＝－146 kPa, onlymarginally lower than the intact undrained uplift capaci-ty, qu, i.e., without a gap, since reverse end bearinggoverns undrained uplift capacity. The initial stiŠness issimilar to the case without gapping, although the ultimateundrained capacity is mobilised at a lower normaliseddisplacement, w/D¿6z. Figure 10 shows pore pressuresbeneath the base plate and indicate similar levels of pas-sive suction are developed with and without gapping(Test T3_1 and T1 respectively).

The undrained uplift capacity following formation of agap and a period of consolidation to allow substantialdissipation of excess pore pressures, qu gap*, was evaluatedin test T4_1. The ultimate undrained capacity is substan-tially reduced compared to the uplift capacity mobilisedeither without a gap or immediately following gap forma-tion, with qu gap*＝－90 kPa. Ultimate undrained capacitywas mobilised at a normalised displacement w/D¿1z,after which the resistance remained relatively constantwith increased displacement, indicating shearing alongthe skirt-soil interface as a mode of failure. The test was

stopped at a normalised displacement w/D¿4z tominimise damage as the site was to be used for a subse-quent test.

The undrained uplift capacity following formation of agap and a period of consolidation (test T4_1) is equiva-lent to 60z of the undrained capacity without gapping,or immediately following gap formation, indicating agreater impact of the gap presence after a time delay.During this period, it appears that softening of the soil inthe gap zone, or further propagation of the gap aroundthe foundation, may have led to increased hydraulic con-nection between the foundation base and the free surface,thus limiting the development of a reverse end bearingmechanism. Figure 10 shows the response of sensors un-derside the base plate and indicates that only about 40zof the negative excess pore pressure measured in the testswithout gapping was generated.

The evolution of the radial stresses during an un-drained uplift stage after large lateral displacements anda previous history of an initial undrained uplift and sus-tained loading stages (test T3_4) is presented in Fig. 11.The reduction of radial stresses on the leading edge, start-ing from a passive condition immediately after the for-mation of the gap, is clear. Although the loss of embed-ment is a contributing factor for this decrease, shearingalong the interface in a softened soil decreases the allowa-ble side friction considerably.

732

Fig. 11. Radial stresses on skirts during transient uplift

Fig. 12. Time histories of time-dependant displacements during sus-tained uplift tests


Sustained Load ResponseSustained uplift loads were deˆned as a proportion of

the undrained uplift capacity for a corresponding ex-perimental condition, i.e., for the case without gapping(test T2), the sustained uplift loads were deˆned as aproportion of the undrained uplift capacity, qu, measuredin test T1. For the cases with gapping, with and withoutconsolidation prior to uplift (tests T3 and T4), the sus-tained uplift loads were deˆned as a proportion of therespective undrained uplift capacity qugap and qugap*.

Sustained uplift loads of 22 and 70z of the corre-sponding respective undrained uplift capacities were con-sidered. It should be noted that the absolute magnitudeof sustained uplift in T3 and T4 corresponding to q/qu＝0.22 and 0.7 are dissimilar owing to the diŠerent un-drained capacity for each case. For the case of sustaineduplift loading applied immediately after the gap forma-tion, q/qu gap＝0.22 and 0.7 correspond to similar magni-tudes in terms of absolute loads (q/qu＝0.21 and 0.69 re-spectively) as the undrained uplift capacity was only mar-ginally aŠected by the presence of the gap. However, forthe case with consolidation following formation of thegap and prior to uplift, q/qugap*＝0.22 and 0.7 correspondto absolute uplift loads of q/qu＝0.13 and 0.42 respec-tively.

To optimise the available space in the sample, the twosustained uplift tests for each case were carried out in thesame site. The lower level of sustained uplift was investi-gated ˆrst to minimise damage in and around the test site.The sequence of each test is presented in Table 1. Aftereach stage of the test, the foundation was pushed backunder load control to the consolidation stress, sc, whichwas then maintained until pore pressures that were gener-ated during the previous stage and the reconsolidationhad substantially dissipated. This allowed the foundationto partially restore its initial position and ensure a similareŠective stress state at the start of each test. The tests arelabelled according to the position in the sequence, e.g.,T4_2 corresponds to the second test in the site for test T4.The sustained loading tests were carried out until either anormalised time-dependant displacement of wt/D＝2zor a prototype duration of tp＝1 year was reached,whichever occurred ˆrst. The threshold values were basedon considerations of displacement based design and the

expected durations of uplift for the new concept ofoŠshore storage platforms that would be buoyant follow-ing oŒoading.

Figure 12 shows time histories of consolidation upliftdisplacements normalised by the foundation diameter.The time history shows the time-dependant component ofthe total displacement, wt, that is the displacement occur-ring after the respective stress level was reached (eliminat-ing the immediate displacement). The results for the testswith no gapping (T2_1 and T2_2 for q/qu＝0.22 and 0.7respectively) conˆrm the expected trend of higher dis-placement rates for higher levels of load and indicate thedisplacement rates remain relatively constant for most ofthe duration of the tests. For the case of sustained upliftloading applied immediately following formation of agap (T3_2 and T3_3), where the applied uplift q/qu gap＝0.22 and 0.7 corresponds to similar magnitudes in termsof absolute loads, q/qu＝0.21 and 0.69 respectively,higher displacement rates are observed at both levels ofsustained load compared to the case without gapping, aswould be expected. Terminal displacement rates againremain relatively constant at both levels of sustained up-lift. For the cases of sustained uplift with consolidationfollowing the formation of the gap (T4_2 and T4_3), thetime histories presented in Fig. 12 indicate a reduction inthe rate of displacement compared to the case with upliftimmediately following gap formation. However, the low-er undrained uplift capacity for the case with consolida-tion following gap formation, leads to markedly lowerabsolute uplift loads, q/qu＝0.13 and 0.42 correspondingto q/qugap*＝0.22 and 0.7 respectively masking the truetrend. In terms of absolute load, a consistent trend of in-creased rate of displacement is observed with increasingload. It is interesting to note the relatively similarresponse in T4_3 and T3_3 for considerably diŠerent ab-solute levels of sustained load (q/qu＝0.42 and 0.69, re-spectively). Reduced resistance along the skirt-soil inter-face and propagation of the gap support this observation.

The dashed line in Fig. 12 represents the displacementrate during the undrained uplift tests, and indicates therapid rate of uplift during sustained loading immediatelyfollowing formation of a gap, and the detrimental eŠect

733

Fig. 13. Time histories of total displacements for high-level of sus-tained uplift

Fig. 14. EŠect of gapping on degradation of uplift resistance withtime


of the presence of a gap at high levels of sustained load. Itshould be noted that these sustained uplift tests, at thehigher level of load, were carried out in sites with a stresshistory of previous gap formation and uplift, rather thanfollowing an initial gap in a virgin site, which would haveled to increasing extension of the gap both radially andwith depth.

A detailed response for the higher sustained uplift level(70z of the corresponding undrained capacity) ispresented in Fig. 13 in terms of total normalised displace-ment, w/D, i.e., the immediate and time-dependent com-ponents of displacement. For the case of uplift immedi-ately after the gap formation (T3_3; q/qu＝0.69), a dra-matic change in the displacement rate after a normaliseddisplacement of w/D¿1.3z is observed which occurredbefore reaching the total level of load and may indicatethe impossibility of the foundation to generate highersuctions. In a given situation it could be possible thatfailure occurs before reaching a speciêd level of sus-tained uplift.

Degradation of Uplift CapacityFigure 14 shows the results from the sustained uplift

tests in terms of degradation of uplift capacity with timefor time-dependent displacements wt/D＝0.5, 1 and 2z.The results from the tests described in this paper areshown as discrete data points with sustained uplift loadsnormalised against the undrained capacity for the casewithout a gap, qu. The continuous `backbone' curveswere derived from results from previous tests at load lev-els q/qu＝0.22 and 0.45 carried out with the same founda-tion in similar soil conditions that maintained intact con-tact along the skirt-soil interface (Gourvenec et al., 2008).The data from this series of tests without gapping (T2) forq/qu＝0.22 coincide with the previous observations,while that for q/qu＝0.7 allows reˆnement of the previ-ous backbone curve.

The presence of a gap shifts the degradation curves tothe left, indicating the increased rate of degradation ofuplift capacity with time. The results show that the eŠectof gapping is clearly potentially signiˆcant in terms ofpredicting the uplift resistance of shallowly skirted foun-

dations. For example, under a sustained uplift load ofq/qu＝0.22, an uplift displacement of wt/D＝0.01 isreached in approximately 380 days when intact contact ismaintained along the skirt-soil interface and in 34 daysimmediately after formation of a gap. The presence of agap becomes more signiˆcant at higher load level, withfor example, the duration for relative displacements w/D＝0.5z and 2z under an uplift load of q/qu＝0.7 beingrelatively similar for the case with a gap but diŠers by ap-proximately seven-fold without gapping. Linearly inter-polating between the available data points indicates thattrend lines for consolidation following gap formation lieto the left of those for either no gapping or gappingwithout consolidation, indicating the detrimental eŠecton sustained uplift resistance of consolidation in thepresence of a gap.

CONCLUDING REMARKS

Results of beam centrifuge tests on a shallowly embed-ded skirted foundation with a skirt depth to foundationdiameter ratio d/D＝0.3, subjected to transient and sus-tained uplift loading have been presented with referenceto the eŠects of gapping along the skirt-soil interface. Aninstrumented model foundation was installed in lightlyover consolidated kaolin clay and a reliable method wasdeveloped to create a gap in-‰ight in the beam centrifuge.The eŠect of gapping was assessed in relation to tests thatmaintained a perfect seal along the skirt-soil interface.

The undrained uplift capacity immediately after for-mation of a gap was found to be largely unaŠected sincereverse end bearing governs resistance. Displacement rateeŠects might be involved and the initial gapping along theskirt-soil contact could be resealed by adjacent soil pulledinto a reverse end bearing failure mechanism. Converse-ly, consolidation following formation of a gap and priorto uplift had a signiˆcant eŠect, reducing undrained up-lift capacity to 60z of that without a gap (or immediatelyfollowing formation of a gap). The rate of displacementunder sustained uplift was adversely aŠected by thepresence of a gap along the skirt-soil interface, with orwithout consolidation following gap formation. The


eŠect was slightly less signiˆcant for uplift immediatelyfollowing formation of a gap.

In summary, the results from the series of centrifugetests presented in this paper indicate the detrimental eŠectof gapping along a foundation skirt on transient and sus-tained uplift capacity. The ˆndings highlight the necessityto consider carefully the likelihood of gapping occurringin the êld in the design of shallow skirted foundationsystems for oŠshore structures. Complementary founda-tion elements to avoid gap formation (so-called crack ar-restors) should be implemented if intact skirt-soil contactis considered.

ACKNOWLEDGEMENTS

The work described here forms part of the activities ofthe Centre for OŠshore Foundation Systems, establishedunder the Australian Research Council's Research Cen-tres Program, and now supported through grantsFF0561473 and DP0665958 and Centre of Excellencefunding from the State of Western Australia. Thisproject was supported through research grantDP0988904. This support is gratefully acknowledged.The authors would also like to thank Mr Don Herley,Senior Beam Technician, and Mr Shane De Catania,Electronics Technician, for their assistance during thetests.

NOTATION

dm, dp: skirt depth of foundation (in model and proto-type dimensions)

Dm, Dp: diameter of foundation (in model and prototypedimensions)

NT-bar: T-bar factorq: average vertical stress under foundation

qu: undrained uplift capacity for ultimate limit state(ULS)

qu gap: undrained uplift capacity immediately followinggap formation

qu gap*: undrained uplift capacity with a period of con-solidation following gap formation

Q: vertical loadtp: time in prototype units[: rate of displacement (T-bar and foundation

tests)V: normalised velocity group ([L/cv)w: total uplift displacementwt: time-dependant uplift displacementwH: total lateral displacement of foundation

z: prototype depth (at skirt tip level when referringto foundation)

sc: stress level during consolidationsi: installation resistance at touch-down

REFERENCES

1) Acosta-Martinez, H. E. and Gourvenec, S. M. (2006): One-dimen-sional consolidation tests on kaolin clay, Report Research Report

GEO: 06385, Centre for OŠshore Foundations Systems, School ofCivil and Resource Engineering, The University of Western Austra-lia.

2) Acosta-Martinez, H. E. and Gourvenec, S. M. (2008): Response ofskirted foundations for buoyant facilities subjected to cyclic upliftloading, Proc. 18th Intl. OŠshore and Polar Engineering Conf.Vancouver, BC, Canada. July 6–11, 2008. ISOPE, 2, 705–712.

3) Acosta-Martinez, H. E. and Gourvenec, S. (2010): Installationresistance and bearing capacity of a shallow skirted foundation inclay, Proc. 7th International Conference on Physical Modelling inGeotechnics. Balkema, Zurich, 2, 1011–1017.

4) Acosta-Martinez, H. E., Gourvenec, S. M. and Randolph, M. F.(2008a): An experimental investigation of a shallow skirted founda-tion under compression and tension, Soils and Foundations, 48(2),247–254.

5) American Petroleum Institute (API) (2000): RP2A: Recommendedpractice for planning, designing and constructing ˆxed oŠshoreplatforms—Working Stress Design, Washington, DC.

6) Andersen, K. H. and Jostad, H. P. (1999): Foundation design ofskirted foundations and anchors in clay, In OŠshore TechnologyConference. Houston, USA, Paper OTC 10824.

7) Bye, A., Erbrich, C., Rognlien, B. and Tjelta, T. I. (1995): Geo-technical design of bucket foundations, In 27th Annual OŠshoreTechnology Conference, Houston, OŠshore Technology Confer-ence, Paper OTC 7793.

8) Clukey, E. C., Aubeny, C. P. and MurŠ, J. D. (2003): Comparisonof analytical and centrifuge model tests for suction caissons sub-jected to combined loads, In International Conference on OŠshoreMechanics and Artic Engineering, OMAE2003-37503.

9) Clukey, E. C., Morrison, M. J., Garnier, J. and Cort áe, J. F. (1995):The response of suction caissons in normally consolidated clays tocyclic TLP loading conditions, In 27th Annual OŠshore Technolo-gy Conference. Houston, Texas, USA. OŠshore Technology Con-ference, Paper OTC 7796.

10) CoŠman, R. A., El-Sherbiny, R. M., Rauch, A. F. and Olson, R.E. (2004): Measured horizontal capacity of suction caissons, InOŠshore Technology Conference. Houston, OTC 16161.

11) Det Norske Veritas (DNV) (1992): Classiˆcation notes No. 30.4,Foundations.

12) Dyvik, R., Andersen, K. H., Hansen, S. B. and Christophersen, H.P. (1993): Field tests of anchos ir clay, I: Description. Journal ofGeotechnical Engineering Division, 119(10), 1515–1531.

13) Finnie, I. M. S. and Randolph, M. F. (1994): Punch-through andliquefaction induced failure of shallow foundations in calcareoussediments, Proc. 7th International Conference on the Behaviour ofOŠshore Structures, BOSS'94. eds. by C. Chryssostomidis, M. S.Triantafyllou, A. J. Whitle and M. S. Hoo Fatt, Boston. Perga-mon, 1: Geotechnics, 217–230.

14) Garnier, J., Gaudin, C., Springman, S. M., Culligan, P. J., Good-ings, D., Konig, D., Kutter, B., Phillips, R., Randolph, M. F. andThorel, L. (2007): Catalogue of scaling laws and similitude ques-tions in geotechnical centrifuge modelling, International JournalPhysicall Modelling in Geotechnics, 7(3), 1–23.

15) Gourvenec, S., Acosta-Martinez, H. E. and Randolph, M. F.(2007): Centrifuge model testing of skirted foundations for oŠshoreoil and gas facilities, 6th International OŠshore Site Investigationand Geotechnics Conference: Confronting New Challenges andSharing Knowledge. London. Society for Underwater Technology(SUT), 479–484.

16) Gourvenec, S. M., Acosta-Martinez, H. E. and Randolph, M. F.(2008): Experimental study of uplift resistance of shallow skirtedfoundations in clay under transient and sustained concentric load-ing. G áeotechnique, 59(6), 525–537.

17) International Orgauization for Standurdization (ISO) (2003):Petroleum and natural gas industries: OŠshore structures: Part 4:Geotechnical and Foundation Design considerations. In Interna-tional Organisation for Standardisation 19901-4.

18) Muir Wood, D. (2004): Geotechnical Modelling, Spon Press, Lon-don.

19) Randolph, M. F. and House, A. R. (2002): Analysis of suction cais-


son capacity in clay, In Proc. 2002 OŠshore Technology Confer-ence. Houston, Texas, Paper OTC 14236.

20) Randolph, M. F. and Hope, S. (2004): EŠect of cone velocity oncone resistance and excess pore pressures, In Proc. EngineeringPractice and Performance of Soft Deposits, IS-OSAKA 2004, Osa-ka, 147–152.

21) Randolph, M. F., O'Neill, M. P. and Erbrich, C. (1998): Perfor-mance of suction anchors in ˆne-grained calcareous soils, In An-nual OŠshore Technology Conference. Houston, 521–529.

22) Randolph, M. F., Jewell, R. J., Stone, K. J. L. and Brown, T. A.(1991): Establishing a new centrifuge facility, In Proc. Internation-al Conference Centrifuge 91. eds. by H.-Y. Ko and F. G. McLean,Boulder, Colorado. A.A. Balkema, 3–9.

23) Stewart, D. P. and Randolph, M. F. (1991): A new site investiga-tion tool for the centrifuge, In Proc. International Conference Cen-trifuge 91. eds. by H.-Y. Ko and F. G. McLean, Boulder, Colora-do. A.A. Balkema, 531–538.

24) Stewart, D. P. and Randolph, M. F. (1994): T-bar penetration test-

ing in soft clay, Journal of Geotechnical Engineering, 120(12),2230–2235.

25) Støve, O. J., Bysveen, S. and Christophersen, H. P. (1992): Newfoundation systems for the Snorre development, In Proc. AnnualOŠshore Technology Conference. Houston, Paper OTC 6882.

26) Supachawarote, C., Randolph, M. F. and Gourvenec, S. (2005):The eŠect of crack formation on the inclined pull-out capacity ofsuction caissons, Proc. 11th International Association for Com-puter Methods and Advances in Geomechanics Conference, Turin,Italy. Balkema, 577–584.

27) Taylor, R. N. (1995): Centrifuges in modelling: Principles and scaleeŠects, Geotechnical Centrifuge Technology, Blackie Academic &Professional, Glasgow, Scotland. 19–33.

28) Tjelta, T. I. and Haaland, G. (1993): Novel foundation concept fora jacket ˆnding its place, In OŠshore Site Investigation and Foun-dation Behaviour, Society for Underwater Technology, 28,717–728.

Documents

EFFECT OF GAPPING ON THE TRANSIENT AND SUSTAINED …